Automation Templates for n8n, Kestra and Apache Airflow

Lead router

Contact-form submissions land in an inbox nobody owns, so the urgent ones wait behind the rest.

1. Trigger

   A POST to an n8n webhook from your contact form.

2. 1

   Step 1

   Normalize the submission into a flat record with a timestamp

3. 2

   Step 2

   Branch on the enquiry type

4. 3

   Step 3

   Post consultation requests to a priority Slack channel, everything else to the general inbox

5. 4

   Step 4

   Email the sender an acknowledgement, then answer the webhook

6. Outcome

   Every enquiry is visible in Slack within seconds, sorted, and the sender knows it arrived.

Trigger

A POST to an n8n webhook from your contact form.

Steps

1. 1.Normalize the submission into a flat record with a timestamp
2. 2.Branch on the enquiry type
3. 3.Post consultation requests to a priority Slack channel, everything else to the general inbox
4. 4.Email the sender an acknowledgement, then answer the webhook

Outcome

Every enquiry is visible in Slack within seconds, sorted, and the sender knows it arrived.

You supply

• Slack credential with chat:write
• SMTP credential for the acknowledgement email

Pipeline failure digest

CI notifications are either off, or so constant that nobody reads them.

1. Trigger

   A schedule, every weekday at 09:00.

2. 1

   Step 1

   Fetch the last 24 hours of GitLab CI pipelines for one project

3. 2

   Step 2

   Count them by status and compute a pass rate

4. 3

   Step 3

   Branch on whether anything failed

5. 4

   Step 4

   Post a digest with the failing branches and their URLs, or stay silent

6. Outcome

   One message on the mornings that need one, and nothing on the mornings that do not.

Trigger

A schedule, every weekday at 09:00.

Steps

1. 1.Fetch the last 24 hours of GitLab CI pipelines for one project
2. 2.Count them by status and compute a pass rate
3. 3.Branch on whether anything failed
4. 4.Post a digest with the failing branches and their URLs, or stay silent

Outcome

One message on the mornings that need one, and nothing on the mornings that do not.

You supply

• GitLab API credential with read_api
• Slack credential with chat:write

Nightly Postgres rollup

Reporting queries run against the live orders table at whatever hour someone opens a dashboard.

1. Trigger

   A schedule, 02:00 UTC every night.

2. 1

   Step 1

   Aggregate yesterday's orders per customer in SQL, where the grouping is cheapest

3. 2

   Step 2

   Render the result to a CSV in a Python task and log the headline numbers

4. 3

   Step 3

   Expose the CSV as a flow output anyone can download from the execution

5. 4

   Step 4

   Alert Slack if any step fails

6. Outcome

   A dated rollup file every morning, and one alert instead of a silent gap when it breaks.

Trigger

A schedule, 02:00 UTC every night.

Steps

1. 1.Aggregate yesterday's orders per customer in SQL, where the grouping is cheapest
2. 2.Render the result to a CSV in a Python task and log the headline numbers
3. 3.Expose the CSV as a flow output anyone can download from the execution
4. 4.Alert Slack if any step fails

Outcome

A dated rollup file every morning, and one alert instead of a silent gap when it breaks.

You supply

• POSTGRES_URL, POSTGRES_USERNAME and POSTGRES_PASSWORD secrets
• SLACK_WEBHOOK secret
• A Docker task runner for the Python step

API uptime and latency check

An endpoint degrades on a Friday evening and the first report comes from a customer on Monday.

1. Trigger

   A schedule, hourly.

2. 1

   Step 1

   Call the endpoint with the latency budget enforced as the task timeout

3. 2

   Step 2

   Compare the status code against the expected one

4. 3

   Step 3

   Log a healthy result, or alert Slack and fail the execution

5. 4

   Step 4

   Catch a timeout or an unreachable host in the error handler and alert on that too

6. Outcome

   A degraded endpoint pages you within the hour instead of within the week.

Trigger

A schedule, hourly.

Steps

1. 1.Call the endpoint with the latency budget enforced as the task timeout
2. 2.Compare the status code against the expected one
3. 3.Log a healthy result, or alert Slack and fail the execution
4. 4.Catch a timeout or an unreachable host in the error handler and alert on that too

Outcome

A degraded endpoint pages you within the hour instead of within the week.

You supply

• SLACK_WEBHOOK secret
• Network access from the Kestra worker

Daily metrics rollup

A reporting job written as a cron script has no retries, no history and no way to backfill a missed day.

1. Trigger

   A DAG schedule, 03:00 UTC daily.

2. 1

   Step 1

   Extract the interval's raw events from an internal HTTP API

3. 2

   Step 2

   Group them per channel and round the totals

4. 3

   Step 3

   Publish the summary to a reporting endpoint

5. 4

   Step 4

   Retry twice on a transient failure before the run goes red

6. Outcome

   A rollup with a run history, a retry policy and a backfill you can trigger for any past day.

Trigger

A DAG schedule, 03:00 UTC daily.

Steps

1. 1.Extract the interval's raw events from an internal HTTP API
2. 2.Group them per channel and round the totals
3. 3.Publish the summary to a reporting endpoint
4. 4.Retry twice on a transient failure before the run goes red

Outcome

A rollup with a run history, a retry policy and a backfill you can trigger for any past day.

You supply

• metrics_source_url, metrics_sink_url and metrics_api_token Airflow Variables
• requests on the worker, no provider packages

Regression suite and report

The full regression suite is too slow for every commit, so it stops running at all.

1. Trigger

   A DAG schedule, weekday mornings at 01:00 UTC.

2. 1

   Step 1

   Run the suite in its checkout and keep the JUnit XML it writes

3. 2

   Step 2

   Sum failures, errors and skips across every test suite in the report

4. 3

   Step 3

   Publish a one-line summary to the task log and the run output

5. 4

   Step 4

   Post to an alert webhook only when the suite is red

6. Outcome

   The slow suite runs every weekday, and only interrupts anyone when it finds something.

Trigger

A DAG schedule, weekday mornings at 01:00 UTC.

Steps

1. 1.Run the suite in its checkout and keep the JUnit XML it writes
2. 2.Sum failures, errors and skips across every test suite in the report
3. 3.Publish a one-line summary to the task log and the run output
4. 4.Post to an alert webhook only when the suite is red

Outcome

The slow suite runs every weekday, and only interrupts anyone when it finds something.

You supply

• regression_working_dir, regression_command and regression_alert_webhook Airflow Variables
• A worker that can run the suite: checkout, interpreter and dependencies